How does income inequality and its structure affect credit? We extend the theoretical framework by Kumhof et al. (2015) to distinguish between upper, middle and low-income classes, and show that most of the positive impact of inequality on credit predicted by Kumhof et al. (2015) should be driven by the share of total output owned by the middle classes. Consistently, this impact should weaken in countries where financial markets are insufficiently developed. These theoretical predictions are empirically confirmed by a study based on a 41-country dataset over the period 1970-2014, where exogenous variations of inequality are identified with a new instrument variable, the total number of ILO conventions signed at the country-level.